With the continuous emergence of cyber-attacks,traditional intrusion detection methods become increasingly limited.In the field of network security,new intrusion detection methods are needed to ensure network security...With the continuous emergence of cyber-attacks,traditional intrusion detection methods become increasingly limited.In the field of network security,new intrusion detection methods are needed to ensure network security.To solve the problems,relevant knowledge involvedwas first introduced.Then,an intrusion detection system based on neural network was designed according to the general intrusion detection framework,and the design of the event collector and analyzer in the systemwas described in detail.Experimentswere conductedwith theweight initialization method of the neural network model,the selection of the activation function,and the selection of the optimizer.Finally,the most suitable hyperparameters were determined and the optimal neural network model was trained.The test results show that the application of neural network to the intrusion detection system can greatly improve the accuracy of intrusion detection,thereby improving the security of computer networks.展开更多
基金National Natural Science Foundation of China(62072416)Henan Province Science and Technology Research Project(202102210176)+2 种基金Zhongyuan Science and Technology Innovation Leader of Zhongyuan Talent Project(214200510026)Henan Province Science and Technology Project(212102210429)The fourth batch of innovative leading talents of Zhihui Zhengzhou 1125 talent gathering plan(ZhengZheng[2019]No.21).
文摘With the continuous emergence of cyber-attacks,traditional intrusion detection methods become increasingly limited.In the field of network security,new intrusion detection methods are needed to ensure network security.To solve the problems,relevant knowledge involvedwas first introduced.Then,an intrusion detection system based on neural network was designed according to the general intrusion detection framework,and the design of the event collector and analyzer in the systemwas described in detail.Experimentswere conductedwith theweight initialization method of the neural network model,the selection of the activation function,and the selection of the optimizer.Finally,the most suitable hyperparameters were determined and the optimal neural network model was trained.The test results show that the application of neural network to the intrusion detection system can greatly improve the accuracy of intrusion detection,thereby improving the security of computer networks.
